RESEARCH RESOURCE PROTEOMICS A systems-wide screen identifies substrates of the SCFbTrCP ubiquitin ligase Teck Yew Low,1,2 Mao Peng,1,2* Roberto Magliozzi,3* Shabaz Mohammed,1,2† Daniele Guardavaccaro,3 Albert J. R. Heck1,2‡ Cellular proteins are degraded by the ubiquitin-proteasome system (UPS) in a precise and timely fashion. Such precision is conferred by the high substrate specificity of ubiquitin ligases. Identification of substrates of ubiquitin ligases is crucial not only to unravel the molecular mechanisms by which the UPS controls protein degradation but also for drug discovery purposes because many established UPS substrates are implicated in disease. We developed a combined bioinformatics and affinity purification– mass spectrometry (AP-MS) workflow for the system-wide identification of substrates of SCFbTrCP,a member of the SCF family of ubiquitin ligases. These ubiquitin ligases are characterized by a multi- subunit architecture typically consisting of the invariable subunits Rbx1, Cul1, and Skp1, and one of 69 F-box proteins. The F-box protein of this member of the family is bTrCP. SCFbTrCP binds, through the WD40 b repeats of TrCP, to the DpSGXX(X)pS diphosphorylated motif in its substrates. We recovered 27 pre- Downloaded from viously reported SCFbTrCP substrates, of which 22 were verified by two independent statistical proto- cols, thereby confirming the reliability of this approach. In addition to known substrates, we identified 221 proteins that contained the DpSGXX(X)pS motif and also interacted specifically with the WD40 repeats of bTrCP. Thus, with SCFbTrCP, as the example, we showed that integration of structural infor- mation, AP-MS, and degron motif mining constitutes an effective method to screen for substrates of ubiquitin ligases. http://stke.sciencemag.org/ INTRODUCTION example, an E3 ubiquitin ligase can be perturbed by RNAi (RNA inter- Relative to most other protein posttranslational modifications (PTMs), ference), chemical inhibitors, or genetic manipulations, whereafter the per- ubiquitylation is unique because ubiquitin can be conjugated to a substrate turbed samples and controls may be analyzed by quantitative mass protein in various chain linkages through which the modified substrates spectrometry (MS) for any change in protein abundance or the amount are sorted to different cellular fates (1). Substrates polyubiquitylated with of PTM-modified proteins (8–10). The availability of an antibody that rec- Lys48 linkages are recognized and degraded by the 26S proteasome (2). ognizes diglycine residues, derived from ubiquitin after trypsin cleavage Target substrates are modified by the sequential actions of a ubiquitin- (11), enables the isolation and identification by MS of ubiquitylated pep- on May 9, 2018 activating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and a ubiqui- tides, thereby providing information about the site of ubiquitylation to tin ligase (E3). The huge repertoire of more than 700 E3 ubiquitin ligases complement quantitative assays (12, 13). in mammals provides the specificity required for the recognition of diverse Alternative approaches based on affinity purification followed by mass substrates (3, 4). Because numerous substrates targeted by the ubiquitin- spectrometry (AP-MS) (14–16) using epitope tagging are also feasible for proteasome system (UPS) are oncoproteins or tumor suppressors, which such screens. In AP-MS methods, a protein of interest (bait) is fused in are potential targets for cancer therapy (5, 6), it is crucial to develop high- frame with an epitope tag, which enables the isolation of the protein com- throughput assays to identify the complete repertoire of substrates of spe- plexes (preys). The affinity-purified complexes are then identified by MS. cific E3 ubiquitin ligases. Although AP-MS experiments have contributed to the identification of Because E3 ubiquitin ligases, as well as other modifying enzymes, are substrates of monomeric E3s, identification of substrates of multisubunit believed to interact transiently with their substrates in a “kiss-and-run” E3s is hampered by the weak affinity of substrates for the subunit of the manner (7), researchers tend to assay the products of the enzymatic activ- E3 that recognizes the substrate and because the complex between the sub- ity rather than the enzyme-substrate interactions. To this end, quantitative strate and the recognition subunit is transient (17). proteomics methods are widely used because they allow the identification Here, we used a stringent AP-MS approach to systematically identify and quantitation of protein substrates with relatively high throughput. For substrates of SCFbTrCP, a ubiquitin ligase that targets key regulators of cell proliferation, cell growth (15), and apoptosis (18) for proteasomal degra- dation. SCFbTrCP consists of Skp1, Cul1, Rbx1, and the F-box protein 1Biomolecular Mass Spectrometry and Proteomics, Bijvoet Center for Biomo- bTrCP that acts as a substrate recognition subunit (Fig. 1). Mammals have lecular Research and Utrecht Institute for Pharmaceutical Sciences, Utrecht Uni- two paralogs of bTrCP: bTrCP1 (also known as FBXW1) and bTrCP2 versity, Padualaan 8, 3584 CH Utrecht, Netherlands. 2Netherlands Proteomics 3 – (also known as FBXW11); however, their biochemical properties appear Center, Padualaan 8, 3584 CH Utrecht, Netherlands. Hubrecht Institute KNAW bTrCP (Royal Netherlands Academy of Arts and Sciences) and University Medical indistinguishable. Although more than 50 substrates of SCF have Center Utrecht, Uppsalalaan 8, 3584 CT Utrecht, Netherlands. been identified since 1999, the list is undoubtedly incomplete. Certain *These authors contributed equally to this work. E3 ligases of the SCF family recognize their substrates through a phos- †Present address: Departments of Chemistry and Biochemistry, University of bTrCP Oxford, New Biochemistry Building, South Parks Road, Oxford OX1 3QU, phorylated motif called a phosphodegron. Most known SCF sub- UK. strates share the canonical phosphodegron DpSGXX(X)pS, which is a ‡Corresponding author. E-mail: [email protected] diphosphorylated motif that mediates the interaction of the substrate with www.SCIENCESIGNALING.org 16 December 2014 Vol 7 Issue 356 rs8 1 RESEARCH RESOURCE Ub Ub These sequences included 45 of the 47 reported phosphodegrons used for Phosphodegron Ub Ub Ub generating the consensus sequence; the two that were not recovered were Ub Ub viral protein U (Vpu) and the RNA binding protein ELAV1. Vpu is a viral Substrate DSGXX(X)S P P protein encoded by HIV-1. Because the human Swiss-Prot database that we used for FIMO analysis does not contain this viral protein, the phos- F-box βTrCP Substrate receptor phodegron for Vpu (RAEDSGNESE) was not recovered. Our recalcula- E2 − Adaptor SKP1 WD40 repeats P 6 RBX1 Ub tion shows that the FIMO value for this phosphodegron is 1.9 × 10 . The phosphodegron sequence (TNYEEAAMAI) of ELAV1 is so devi- Scaffold CUL1 N ated from the consensus that the FIMO P value is 0.00124, lower than the − E1 Ub P ≤ 1×103 cutoff, and hence is also not included. Ub Because FIMO P valuescanbeusedtoestimatehowsimilareachpu- Fig. 1. The architecture of wild-type SCFbTrCP. SCFbTrCP consists of the core tative phosphodegron is to the consensus motif, we ranked each matched subunits Cul1, Rbx1, and Skp1, and the F-box protein bTrCP that deter- motif according to its respective FIMO P values (lowest first, represent- mines substrate specificity. Cul1 is a scaffold protein that interacts through ing the closest match to the consensus motif). As expected, the top-ranking its N terminus with the adaptor protein Skp1 and through its C terminus with reported phosphodegrons generally contained motifs that matched more the RING finger protein Rbx1, which in turn binds to a specific ubiquitin- closely to the DpSGXX(X)pS motif, for example, eEF2K (ranked 1), CTNNB1 conjugating enzyme (E2). Skp1 interacts with the F-box domain of bTrCP, (ranked 11), DLG1 (ranked 246), and SIPA1L1 (ranked 404). For ease of which, in turn, recruits substrates through its WD40 repeats. These repeats reference, we named them category I motifs. Following the SIPA1L1, we form a b-propeller structure, which specifically recognizes a diphosphory- observed a sharp drop in the ranking for motifs derived from substrates lated motif with the consensus DpSGXX(X)pS, known as phosphodegron. such as PLK4 (ranked 1214), MYC (ranked 1426), CDC25B (ranked Downloaded from 2931), TWST1 (ranked 3778), and MCL1 (ranked 5218), which have less canonical motifs. The second group is termed category II motifs. the WD40 repeats of bTrCP (Fig. 1) (19, 20). Mutation of either the phos- To further refine and corroborate our motif analysis, we queried the phodegron or the WD40 repeats abolishes substrate binding to bTrCP (21). PhosphoSitePlus repository (35), which contains phosphoproteomics and The phosphodegrons of all reported SCFbTrCP substrates are conserved ubiquitomics data. In total, we found 3881 phosphorylation sites reported in vertebrates (22). for our putative degrons. This analysis revealed that 46% (3507 of 7623) http://stke.sciencemag.org/ A challenge in AP-MS analyses is discriminating bona fide interacting of the proteins with putative degrons have at least one ubiquitylation site partners from nonspecific protein contaminants (23). To resolve this pro- (table S2). However, because of the low abundance of some modifications blem, we used label-free quantitative proteomics in combination with two at specific sites and limitation of